We have a new machine and when the rest of that arrives I'll explore
Exim 4 in depth. Meanwhile, on our Exim 3.36 system, we have some
addresses we expand by alias into lists of people. I shall call the
address in question xyz@???, for the sake of this argument.
Spammers have got hold of this address, and are sending us the usual
fare with senders and recipients set to xyz@???.
So we try
sender_reject_recipients = @@lsearch*;/usr/local/exim/sender_reject_recipients_data
with an entry:
dmu.ac.uk: xyz
This works, but the bounce messages come back to xyz, and get
expanded, so everyone gets the bounces.
I tried adding a director before the system_aliases director which
used an alias file with an
xyz: :blackhole:
entry, only acted upon if the sender was xyz. Then I retried it
to be acted upon when the sender was <>.
However, I could not get these to work.
So I try to continue rejecting early and use the simpler solution:
sender_reject = @@lsearch:/usr/local/exim/blackhole_senders
because mail should never come *from* this list.
blackhole_senders is
---8<---
de-montfort.ac.uk xyz
dmu.ac.uk xyz
eng.cse.dmu.ac.uk xyz
brains.eng.cse.dmu.ac.uk xyz
---8<---
but I'm still getting bounces to the list.
Looking at the bounces I am still getting the bodies, so
presumably these are being processed, i.e, the rejection is not
happening as early as I would expect.
The next thing I can do is hack the system filter to reject things
with from lines matching xyz@???.... But I'll be accepting the
DATA then.
Am I even "warm", or am I barking up the wrong tree completely, to
mix metaphors :-)?
Hugh
